Research Article
Product Sentiment Analysis Using Particle Swarm Optimization Based Feature Selection in a Large-Scale Cloud
@INPROCEEDINGS{10.4108/eai.7-6-2021.2308639, author={P. Vasudevan and K. P. Kaliyamurthie}, title={Product Sentiment Analysis Using Particle Swarm Optimization Based Feature Selection in a Large-Scale Cloud}, proceedings={Proceedings of the First International Conference on Computing, Communication and Control System, I3CAC 2021, 7-8 June 2021, Bharath University, Chennai, India}, publisher={EAI}, proceedings_a={I3CAC}, year={2021}, month={6}, keywords={sentiment analysis feature selection (fs) swarm intelligence (si) np-hard particle swarm optimization (pso)}, doi={10.4108/eai.7-6-2021.2308639} }
- P. Vasudevan
K. P. Kaliyamurthie
Year: 2021
Product Sentiment Analysis Using Particle Swarm Optimization Based Feature Selection in a Large-Scale Cloud
I3CAC
EAI
DOI: 10.4108/eai.7-6-2021.2308639
Abstract
Cloud computing is evolving by shifting its services out of the applications of the firewall, storage, services, and applications that are accessible on the web. After this, the services will be used with the help of the Internet and paid according to the user’s/customer’s needs. Big data can be efficiently and economically analysed using Cloud computing. Sentiment Analysis deals with study of opinions, and is based on the emotions, attitudes, and opinions of this entity. The objective of the proposed work sentiment analysis using Particle Swarm Optimization (PSO) algorithm based on new Feature Selection (FS) method. FS method is quite complex and a demanding task in terms of computation more so for a high dimension dataset. Swarm Intelligence (SI) is techniques capable of solving computational problems that are NP-hard (Non-Deterministic Polynomial time). It is gaining plenty of popularity to solve the problems of optimization and FS. Particle Swarm Optimization (PSO) is used widely for solving problems of optimization and also the problems of FS. The Support Vector Machine (SVM) analyses data and further identify patterns utilized for classification purposes. In this work, a PSO-based FS is proposed for product sentiment analysis. The classification accuracy achieved by PSO based FS is higher by 5.93% and by 6.91% for 20% training when compared to IG and GA based FS, respectively. Similarly, classification accuracy achieved by PSO based FS is hi